PERSPECTIVES
OPINION
Proteostasis in complex dendrites
Cyril Hanus and Erin M. Schuman
Abstract | Like all cells, neurons are made of proteins that have characteristic
synthesis and degradation profiles. Unlike other cells, however, neurons have a
unique multipolar architecture that makes ~10,000 synaptic contacts with other
neurons. Both the stability and modifiability of the neuronal proteome are crucial
for its information-processing, storage and plastic properties. The cell biological
mechanisms that synthesize, modify, deliver and degrade dendritic and synaptic
proteins are not well understood but appear to reflect unique solutions adapted
to the particular morphology of neurons.
Although it has not been directly measured,
the proteome of a single neuron is certainly
immense, probably including ~10,000 different types of proteins, with total protein numbers predicted to be in the range of 4 × 1010
molecules per cell1–3. The complex morphology of neuronal dendrites and synapses
(FIG. 1), with dendrite arbors that encompass
~50,000 μm and ~5,500 μm3 of length and
volume1,2, respectively, pose unique challenges for the optimization and homeostatic
regulation of protein concentration in space
and time. In dendrites, the machineries for
both protein synthesis and protein degradation are present and have been shown to
regulate protein availability during synaptic
transmission4. In addition, emerging data
indicate that proteins exhibit high mobility
between synapses5 and high concentration
gradients along the length of the dendrite6
(FIG. 1b). As such, the minimal compartment that serves as the elementary unit for
synaptic plasticity (for example, a synapse, a
cluster of synapses, a branchlet, and so on)7
(FIG. 1) needs to be considered in the context
of the distribution of local cell biological
machineries.
Although it is probably fair to say that
all of the major synaptic molecules have
been identified, a dynamic understanding
is lacking. Many aspects of local protein
synthesis boil down to the question of
how limiting resources can be produced
and used locally. However, we still know
surprisingly little about protein synthesis
rates, turnover, mobility and exchange at
the scale of the entire neuron. The role and
regulation of local protein synthesis and
degradation during neuronal development
and synaptic plasticity have been reviewed
elsewhere4,8–11. Here, we discuss emerging
dimensions of dendritic and synapse proteostasis and, more specifically, the following
three questions. First, how does neuron size
affect the partition between somatic and
dendritic biosynthetic machineries (FIG. 2a)?
Second, what are the spatial scales of subcellular compartmentalization in dendrites
(FIG. 2b)? Last, how does this compartmentalization determine the impact of protein
synthesis on local membrane composition
(FIG. 2c)?
The constraints on cell biology
It has long been recognized that the gigantic size and complexity of neuronal protrusions (FIG. 1a) must exert a colossal pressure
on cellular metabolism12,13. With surface
areas up to 10,000 times larger than those
of fibroblasts, neurons probably evolved
specialized means to scale up cellular processes, particularly the production, trafficking and turnover of membrane components
(FIG. 2a). Indeed, there are many examples
of neuron-specific features of common
cellular ‘machines’ in invertebrates and
mammals14–19, which suggests that neurons
solved the grand size problem via qualitative changes rather than just scaling up
existing solutions. Perhaps consistently,
638 | SEPTEMBER 2013 | VOLUME 14
recent attempts to relate brain energy
consumption to neuron numbers and
morphology across phyla suggest that the
scaling up of cortex size during evolution
did increase energetic needs but, regardless
of neuron size, occurred with a fixed energy
budget per neuron20.
To what extent do neuron size and
morphological complexity affect metabolic
needs and basic cellular features such as the
abundance, distribution and morphology
of neuronal organelles? Are there specific
cell dimensions above which those relationships show non-linear scaling (FIG. 2a)? These
questions remain largely unexplored. During
neuronal development, the scaling up of cellular structures with dendritic growth may
be possible only up to a certain point. For
example, polyribosomes occur every 1 to
2 μm along the lengths of both immature
and mature hippocampal dendrites despite
an increased density of excitatory synapses
in more mature dendrites21,22. Differences in
the regulation of basic cellular functions also
exist between different neuronal types and
developmental stages. For example, the morphology of the neuronal Golgi apparatus,
one of the main components of the secretory
pathway, differs from one neuron type to
another, notably among principal excitatory
cells of the hippocampus (FIG. 3). It is known
that the morphology of the Golgi system
directly reflects the specific secretory needs
of a cell23. It is thus likely that distinct dendritic lengths impose specific demands on
secretory functions and thereby affect Golgi
morphology 24. However, the relationship
between dendritic length and the morphology of the somatic Golgi is weakly correlated
(FIG. 3c), suggesting a more complicated
relationship between secretory demands
and the organization of secretory organelles
(C.H., E.M.S., L. Kochen and S. tom Dieck,
unpublished observations). Whether this
reflects distinct and neuron-type-specific
divisions of membrane protein synthesis at
local versus somatic sources remains to be
determined.
Similar to many cost–benefit discussions
of evolutionary problems, the question of
how neuron morphological complexity
emerged while preserving cell biological
functionality is a chicken and egg problem.
www.nature.com/reviews/neuro
© 2013 Macmillan Publishers Limited. All rights reserved
PERSPECTIVES
a
Frog
Lizard
Rat
b
Human
Kv4.2
HCN1
SLM
SR
SP
SO
50 μm
c
50 μm
d
0 min
35 min
100 min
1 μm
1 μm
Figure 1 | Size and compartmentalization of the neuronal membrane. a | Pyramidal cell development in vertebrates. The top panel shows the morphological appearance of principal neurons in the
cortex of frogs, lizards, rats and humans. The bottom panel shows the
morphology
pyramidal cells
Nature
Reviewsof
| Neuroscience
at different developmental stages. Note the progressive increase of pyramidal neuron size and morphological complexity (right to left). b | Immunolabelling of a hippocampal slice highlights the contrasting distribution of two different potassium channels (hyperpolarization-activated cyclic
nucleotide-gated channel 1 (HCN1) and Kv4.2) in the apical dendrites of CA1 pyramidal neurons.
Whereas HCN1 is enriched in distal dendrites (upper part of left image), Kv4.2 is more abundant in
more proximal segments of dendrites (middle part of right image). The black lines show the correspondence between panels a and b and indicate the orientation of hippocampal CA1 pyramidal neurons. c | Three-dimensional reconstruction of a segment of a CA1 pyramidal neuron apical dendrite,
showing the variability of dendritic spine morphology. d | Enlargement of a single dendritic spine
(white dot) at 0 min, 35 min and 100 min after plasticity induction by focal glutamate uncaging, reflecting the potentiation of an individual synapse. SLM, stratum lacunosum moleculare; SO, stratum oriens;
SP, stratum pyramidale; SR, stratum radiatum. Part a is adapted from REF. 88 © (1894) C. Reinwald &
Cie. Part b is adapted, with permission, from REF. 6 © (2012) Elsevier. Part c is adapted, with permission,
from REF. 41 © (1999) Oxford University Press. Part d is adapted, with permission, from REF. 7 © (2011)
Elsevier.
On the one hand, the huge dimensions
of neurons are a quantitative challenge
to which cells seem to have responded
by diversifying the functions of existing
molecular systems14–19. On the other hand
(and ignoring, for the moment, dendritic
spines), the existence of remote dendritic
segments isolated by many bifurcations and
situated up to millimetres away from the
cell body probably facilitates the compartmentalization of cellular processes (FIG. 2b),
‘passively’ conferring novel properties to
ancient systems. For example, we recently
showed that the extensive branching of
mature dendrites cumulatively compartmentalizes nascent membrane proteins in
the endoplasmic reticulum (ER)25. Together
with other studies26,27, these results indicate
that although a minimal compartmentalization of secretory trafficking is necessary
for asymmetrical dendritic growth and
initial dendritic branching, the increasing complexity of more mature dendrites
amplifies this very compartmentalization.
An interesting possibility is that at a given
point in their development, neurons reach
a critical size and level of complexity that
qualitatively changes their responses to
specific constraints or stimuli. The efficient
exchange of molecules between the distal
dendritic segments and the soma are also
affected by dendritic morphology. It is thus
likely that the relative impact of global and
local protein processing will be different
depending on the developmental stage and
the neuron type.
NATURE REVIEWS | NEUROSCIENCE
Before discussing how these morphological constraints shape resource allocation
within neurons, let us first consider the
diversity and quantities of synaptic proteins,
their local turnover and their site of production
within complex neurons.
Synaptic composition and turnover
Diversity of synaptic components and their
local turnover. The 10‑year period between
1990 and 2000 witnessed an explosion in the
identification of the molecular players that
populate the synapse. Through advances in
biochemistry, molecular biology and the yeast
two-hybrid system, the identification and
cloning of the major receptors and signalling and scaffolding molecules was accomplished28. The number of molecules that
mediate and regulate synaptic transmission
is large. A recent synthesis of several different
proteomic studies identified 2,788 different
proteins as integral components of excitatory
synapses29. In dendritic spines, however, the
proteins that are most closely associated with
the synapse, physically and functionally, are
those that reside in the postsynaptic density
(PSD), an electron-dense structure that contains ~300 different protein types, including
glutamate receptors as well as scaffolding
and signalling molecules30. The molecular
heterogeneity of individual PSDs has been
explored30,31. An average PSD in the rat forebrain has a diameter of 360 nm and a mass of
1.1 ±.4 GDa, with stoichiometries of specific
proteins ranging from to tens to thousands of
molecules (TABLE 1).
Long-term imaging studies in vivo indicate that the physical structure of individual
synapses can be maintained from days up
to months32,33. Despite this structural stability, the turnover (the transit of molecules in
and out of the synapse) of synaptic proteins
within these structures is fast (on the scale
of minutes). Over the past decade, advanced
live-imaging techniques have revealed the
rapid and continuous exchange of synaptic
components between synaptic and extrasynaptic compartments5. Among these
approaches, fluorescence recovery after
photobleaching (FRAP) has been particularly
useful to compare the synaptic dynamics of
various proteins. By bleaching fluorescently
tagged synaptic proteins in defined spatial
areas, the recovery of fluorescence in the
bleached region provides information about
both the molecules that remain resident (the
‘immobile fraction’) as well as those molecules that move into the bleached region
(the ‘recovered or mobile fraction’). FRAP
has, so far, a limited spatial resolution and,
because it is a bulk approach, can mask the
VOLUME 14 | SEPTEMBER 2013 | 639
© 2013 Macmillan Publishers Limited. All rights reserved
PERSPECTIVES
Scaling of biosynthetic machinery with neuronal size
Ab
Biosynthetic capacity
Aa
100%
Dendrite
Soma
0
Small
Large
Neuron size
Spatial scales of dendritic compartmentalization
Bb
Protein levels
Ba
Retention of proteins
at postsynaptic sites
Local dwell-time
Synaptic proteins
PSD
Spine
Cytoplasmic proteins
ER
ER membrane proteins
1 µm
50 µm
Dendrite
Short
Long
Protein content
(normalized)
Protein synthesis, protein exchanges and local changes of membrane composition
Ca
Cb
Synapse 1
Synapse 2
Synapse 1
2
1
0
0
30
60
Local synthesis
Long
Protein
lifetime
Short
Time
Protein
exchange
Slow
Fast
Protein content
(normalized)
Time after synthesis (min)
Synapse 2
2
1
0
0
30
60
Time after synthesis (min)
Figure 2 | Outstanding questions regarding protein synthesis and exchange in complex dendrites. Hypothetical weights of somatic and dendritic biosynthetic machinery during the developReviews |ofNeuroscience
ment of a pyramidal neuron are represented as bubbles (part Aa)Nature
or percentage
the maximal
biosynthetic capacity (part Ab) as a function of neuron size. Distinct spatial scales of proteins compartmentalization in dendrites are shown in part B. The protein concentration gradient throughout
the apical dendrite of hippocampal pyramidal neurons is shown, in which levels of some proteins (for
example, hyperpolarization-activated cyclic nucleotide-gated channel 1 (HCN1)) tend to increase
with distance from the soma (part Ba). Three examples of subcellular compartmentalization that alter
the dwell-time of proteins are shown (part Bb): the accumulation and/or retention of synaptic proteins
at postsynaptic sites, the passive trapping of cytoplasmic proteins in dendritic spines and the constrained mobility of membrane proteins within the endoplasmic reticulum (ER). Protein turnover, protein lifetime and local changes in membrane composition are shown in part C. Local synthesis of two
hypothetical proteins in a segment of dendrite in which synapse location, protein stability and turnover at synapses (part Ca) affect the variation of their relative levels at synapses over time (part Cb). The
production of a short-lived protein with a fast synaptic turnover rate (purple circle; the gradual fading
of colour represents the transit towards the end of the lifetime of the protein) results in rapid and
compartmentalized changes in protein composition in a dendritic spine (synapse 1). By contrast, the
synthesis of a more stable membrane protein (blue rod) that has a slower turnover rate in the dendritic
shaft results in slower onset and more distributed modifications of synaptic composition (synapse 2).
Note that the internalization and recycling of receptors in the dendritic spine are not represented.
640 | SEPTEMBER 2013 | VOLUME 14
complexity of the molecular interactions
that structure the synapse34. However, the
numerous photobleaching studies that are
available allow a first glimpse at the apparent turnover of many postsynaptic proteins.
As shown in TABLE 1, recovery half-times
of the main synaptic proteins vary over an
order of magnitude and overall are slower
for ion channels than scaffolding proteins
(mean ± SEM = 2.8 ± 0.8 min, n = 7, for transmembrane proteins (GABAA receptor subunit
γ2 excluded) and mean ± SEM = 1 ± 0.2 min,
n = 11, for cytoplasmic proteins; P < 0.05
using the t‑test). Similarly, average recovered
fractions show that, as a family, synaptic
membrane proteins exchange over slower
time periods than cytoplasmic proteins in the
PSD (mean ± SEM = 39 ± 4% of initial fluorescence, n = 8, for membrane proteins and
mean ± SEM = 55 ± 6%, n = 11, for cytoplasmic
proteins; P < 0.05 using the t‑test). As suggested by high-throughput measurements of
protein stability (TABLE 2), the lifespan of an
average protein is (albeit with huge variability) on the order of a few days. Taken together,
this indicates that synaptic proteins cycle in
and out of the synapse over many rounds
and, as shown by single-molecule imaging 5, explore multiple synapses before being
degraded5. Although these data provide valuable insights into the timescales over which
synaptic composition and properties can be
remodelled34,35, they give little information on
how ongoing protein synthesis contributes to
the global trading of synaptic components at
the scale of the entire neuron. The apparent
fast turnover of synaptic components at synapses implies the existence of readily available
reserve pools, but strikingly little is known
of the nature, replenishment mechanism
and local versus global accessibility of these
reserve pools (FIG. 2c).
The dendritic transcriptome and the origin
of synaptic components. From which source
(or sources) are synaptic proteins synthesized? Until recently, the relatively small and
largely heterogeneous cast of dendritic and
axonal mRNAs that had been identified36–38
suggested that local protein synthesis is not
a major source of synaptic components and
may be used only in certain circumstances
during synaptic plasticity. The strikingly
little overlap between available studies36–38,
however, indicated that only a proportion
of the local mRNA population had been
identified and prompted us to use more
sensitive methods to identify the full complement of mRNAs present in synaptic regions.
Through a combination of microdissection,
next-generation sequencing, single-mRNA
www.nature.com/reviews/neuro
© 2013 Macmillan Publishers Limited. All rights reserved
PERSPECTIVES
a
DG
CA1
CA3
50μm
b
10 μm
c
5 μm
Golgi volume (µm3)
d
?
200
CA3
150
100
50
0
DG
0
5,000
CA1
10,000
15,000
Total dendritic length (µm)
Nature
Reviews | Neuroscience
Figure 3 | Dendritic
morphology
and the regulation of cellular machinery. a | Semi-schematic
morphologies of rat hippocampal principal neurons in dentate gyrus (DG), CA1 and CA3.
b | Fluorescent images of DG, CA1 and CA3 principal neurons from 4‑week-old rats after immunolabelling of the Golgi protein GM130 (also known
as GOLGA2) (green). (Nuclei are coloured in blue.)
c | Three-dimensional reconstructions of the
somatic Golgi apparatus in DG, CA1 and CA3
principal neurons, showing the distinct morphologies of the somatic Golgi apparatus (immunolabelling is the same as in part b). d | Volume of the
somatic Golgi (mean ± SEM (n = 10–11 neurons) is
taken from two 4-week-old rats) as a function of
total dendritic length in the same neuron populations (dendritic lengths were taken from
REFS 1,89). Note the non-linearity of Golgi volume
increase in neurons with increasing dendritic
length and the potentially linear scaling of this
parameter in neurons from CA3 and the DG (indicated by the question mark). Part a is adapted,
with permission, from REF. 90 © (2007) Oxford
University Press. Parts b–d are taken from C.H.,
E.M.S., L. Kochen and S. tom Dieck, unpublished
observations.
counting and high-resolution in situ hybridization, we recently identified a surprisingly
large number of previously unrecognized
mRNAs in the CA1 synaptic neuropile; we
obtained a conservative dataset of dendritic
and axonal mRNAs (hereafter referred to
as the neuropile transcriptome) containing
more than 2,500 species39. Together with
previous studies40, we estimated that the
total transcriptome of a typical CA1 pyramidal neuron contains at least ~5,000 genes,
which indicates that a large fraction of its
transcriptome can be detected within dendrites39. The soma constitutes less than 5%
of the somatodendritic volume in a rat CA1
pyramidal neuron1,41, which implies that the
huge majority of the total cellular proteome
functions in neurites, a constraint that would
explain why mRNA localization seems to be
the rule rather than the exception.
Importantly, identified dendritic mRNAs
cover diverse classes of proteins and display
highly variable expression levels and distinct
distribution patterns39 (FIG. 3d). These include
voltage-gated channels, neurotransmitter
receptors, synaptic adhesion and scaffolding proteins, signalling molecules as well
as components and regulators of the protein synthesis and degradation machinery.
More recently, we performed 3ʹ‑end deep
sequencing 42,43 of neuropile mRNAs (E.M.S.,
G. Tushev, I. Cajigas and T. Will, unpublished
observations) to analyse relative expression
levels within the neuropile transcriptome.
We found that mRNAs encoding major synaptic proteins (or at least those included in
TABLE 1) are, on average, within the top 25%
of most abundant species (mean percentile
rank ± SEM = 0.75 ± 0.04 for the 25 genes
shown in TABLE 1). To use an alternative
selection criteria, the 2,285 mRNA species
present in our updated neuropile dataset
were compared on the basis of their inclusion
in relevant gene pathways — ‘glutamatergic
synapse’, ‘GABAergic synapse’, ‘long-term
potentiation’ and ‘long-term depression’ — defined in the Kyoto Encyclopedia
of Genes and Genomes (KEGG)44. This
analysis confirmed that mRNAs encoding
proteins involved in fast synaptic transmission were particularly abundant within the
neuropile transcriptome (mean percentile
rank ± SEM = 0.7 ± 0.04 for the 64 genes
included in the synaptic subgroup and mean
percentile rank ± SEM = 0.5 ± 0.006 for the
other 2,221 genes; P < 10−4 using the Mann–
Whitney U-test). Altogether, these data indicate that many key mediators and regulators
of synaptic signalling and dendritic excitability may be produced at a local rather than
somatic source.
NATURE REVIEWS | NEUROSCIENCE
Local synthesis of ion channels and the
organization of the neuronal secretory
pathway. Despite differences in the numbers and identity of mRNA species found
in neuronal processes, several published
studies have identified mRNAs that encode
membrane proteins36–39, raising fundamental questions regarding secretory processing
in dendrites. In contrast to cytoplasmic
proteins, integral membrane proteins of the
cell surface are synthesized, modified and
trafficked through the various membrane
compartments of the secretory pathway.
Although virtually all of the distinct structures that constitute the secretory pathway
can be detected in some dendrites25,45, how
and to what extent nascent membrane proteins can be processed and trafficked locally
is still a matter of debate. Two features of
secretory processing — namely, the apparent continuity of the somatodendritic ER
and the microtubule-dependent transport
of cargo that is required for sequential processing by the distinct membrane-bound
structures of the secretory system46 — are
particularly puzzling, as both could potentially lead to a large spread of proteins
produced at a local source (that is, in dendrites), thus compromising local production. Nevertheless, evidence suggests that
local membrane protein production can
allow for rapid and individual changes in
synaptic composition47–49.
So how have neurons modified the
secretory machinery to preserve some
degree of specificity for remotely synthesized membrane proteins? We have previously shown that local zones of increased
ER complexity compartmentalize ER cargo
at dendritic branch points and near dendritic spines, facilitating ER export and
local processing at discrete secretory hubs25.
Although it is not yet clear how membrane
cargo transport can be locally restricted
Glossary
Local turnover
The continuous replacement of proteins within a given
cellular structure.
Neuropile
Synaptically dense regions of the CNS, containing mostly
dendrites, axons and glial cells and a small number of
neuron cell bodies.
Secretory pathway
An array of membrane bound structures comprising the
endoplasmic reticulum (ER), the ER–Golgi intermediate
compartment, the Golgi apparatus, the trans-Golgi
network and vesiculotubular carriers that synthesizes and
processes secreted and membrane proteins of the cell
surface.
VOLUME 14 | SEPTEMBER 2013 | 641
© 2013 Macmillan Publishers Limited. All rights reserved
PERSPECTIVES
Table 1 | Comparison of protein number per PSD, apparent turnover, lifetime and presence of its mRNA in dendrites
Protein
(copy number*)
Apparent turnover‡
Reporter
Protein half-life
Assay
t1/2
RF
(min)
Refs PSD in vitro
t1/2 (h)
(antigen)
EPSCs
MK‑801
5
30%
(20 min)
62 13.5
YFP-GluN1
FRAP
6.8
36%
91
§
Lysate
in vitro||
t1/2 (days)
(gene name)
Total brain
in vivo¶
t1/2 (days)
(gene name)
Relative
abundance
of dendritic
mRNA#
(gene name)
Integral membrane proteins
NMDAR GluN1,
GluN2A and GluN2B
subunits
(20)
(GluN1)
1.6
(Grin1)
0.5–0.6
(Grin1)
12.2
(GluN2A)
15.7
(GluN2B)
0.1–0.2
(Grin2a)
1.8
(Grin2b)
0–0.1
(Grin2b)
SEP‑GluA1
FRAP
3
41%
92
YFP‑GluA1
FRAP
2.11
55%
91
1.9
(Gria2)
7.2
(Gria2)
0.9–1
(Gria2)
SEP‑GluA2
FRAP
1.6–
3.7
47–53%
93
2
(Gria3)
8.1
(Gria3)
0.7–0.8
(Gria3)
CACNG2 (also known
as stargazin)
CACNG2-GFP
FRAP
0.5
25%
94
mGluR1, mGluR5
(20)
mGluR1
AMPAR GluA1, GluA2
and GluA3 subunits
(60)
GABAAR
2.1
(Gria1)
0.4–0.5
(Cacng2)
4.4
(mGluR1)
mGluR5‑GFP
FRAP
0.5
54%
95
0.8–0.9
(Grm5)
GABAAR α1V257C
subunit IPSCs
MTSES
5
70%
(30 min)
63
0.8–0.9
(Gabra1)
SEP‑GABAAR β3
subunit
FRAP
2
20%
96
0.9–1
(Gabrb3)
SEP‑GABAAR γ2
subunit
FRAP
0.14
40%
97
0.8–0.9
(Gabrg2)
PSD95‑GFP
FRAP
0.8
20% (4 min)
98 7.5
Cytoplasmic proteins
PSD95 (encoded by
Dlg4)
(300)
(PSD95)
PSD95‑Venus
FRAP
1
35% (7 min)
99
PSD95‑YFP
FRAP
4
46%
91
SHANK1–SHANK3
(150)
GFP‑SHANK2
FRAP
2
41%
(30 min)
100
SAP102 (encoded by
Dlg3)
GFP‑SAP102
FRAP
0.28
80%
SAP97 (encoded by
Dlg1)
(10)
YFP‑αSAP97‑
GFP
FRAP
1.1
40%
HOMER1–HOMER3
(60)
GFP‑HOMER1C
FRAP
1
55% (4 min)
3.7
(Dlg4)
15.3
(Dlg4)
4.1
(Shank2)
0.9–1
(Shank1,
Shank3)
101 6.5
2.1
(Dlg3)
0.5–0.6
(Dlg3)
102
5
(Dlg1)
(SAP102)
98 2.7
(HOMER)
9.9
(Dlg1)
4.8
(Homer1)
0.8–0.9
(Syngap1)
YFP‑SAPAP1
FRAP
1
28%
100
0.6–0.8
(Dlgap3,
Dlgap4)
91 9.2
AKAP5
(20)
CaMKIIα, CaMKIIβ
(5,600)
0.5–0.6
(Dlg1)
0.9–1
(Homer2)
SYNGAP
(360)
SAPAP1 (also known
as GKAP)–SAPAP4
(encoded by Dlgap1–
Dlgap4)
(150)
0.9–1
(Dlg4)
0.6–0.7
(Akap5)
(AKAP5)
GFP-CaMKIIα
FRAP
2.35
80%
91 20
(CaMKII)
642 | SEPTEMBER 2013 | VOLUME 14
3–3.8
(Camk2a,
Camk2b)
6.5–8.5
(Camk2a,
Camk2b)
0.9–1
(Camk2a,
Camk2b)
www.nature.com/reviews/neuro
© 2013 Macmillan Publishers Limited. All rights reserved
PERSPECTIVES
Table 1 (cont.) | Comparison of protein number per PSD, apparent turnover, lifetime and presence of its mRNA in dendrites
Protein
(copy number*)
Apparent turnover‡
Reporter
Assay
Protein half-life
t1/2
RF
(min)
Refs PSD in vitro
t1/2 (h)
(antigen)
§
Lysate
in vitro||
t1/2 (days)
(gene name)
Total brain
in vivo¶
t1/2 (days)
(gene name)
91 7.3
PP1
0.7–1
(Ppp1ca,
Ppp1cb)
(PP1)
α-actinin 2
α-actinin
2-Venus
FRAP
0.7
60%
99
β‑catenin
GFP-β-catenin
FRAP
0.04
60%
Actin
GFP-actin
FRAP
0.4
Gephyrin
Venus-gephyrin FRAP
(Ge2.6)
0.3
Relative
abundance
of dendritic
mRNA#
(gene name)
4.7–6.1
(Actn1, Actn4)
8.5
(Actn4)
0.8–0.9
(Actn1)
103
5
(Ctnnb1)
0.8–0.9
(Ctnnb1)
80%
104
15.3
(Actb)
50%
(10 min)
105
AKAP5, A-kinase anchor protein 5; AMPAR, AMPA receptor; CACNG2, voltage-dependent calcium channel γ2 subunit; CaMKIIα, calcium/calmodulin-dependent
protein kinase II α-subunit; CaMKIIβ, CaMKII β-subunit; Dlg, disks large homologue; Dlgap; disks large-associated protein; EPSCs, excitatory postsynaptic currents;
FRAP, fluorescence recovery after photobleaching; GABAAR, GABA type A receptor; GFP, green fluorescent protein; IPSCs, inhibitory postsynaptic currents; mGluR,
metabotropic glutamate receptor; NMDAR, NMDA receptor; PP1, protein phosphatase 1; PSD95, postsynaptic density protein 95; SAP, synapse-associated protein;
SAPAP; SAP90/PSD95-associated protein; SEP, super-ecliptic fluorescent protein; SHANK, SH3 and multiple ankyrin repeat domains protein; t1/2, half-life or
half-time; YFP, yellow fluorescent protein. *Copy numbers per synapse were taken from REF. 28. ‡Apparent turnover half-times and recovery fractions (RFs) at
synapses in vitro. Indicated values were taken, or approximated by us graphically, from the indicated references. Numbers in brackets after recovered fractions
indicate the time at which recovered fractions were quantified when recovery plots did not reach a plateau. §Protein stability half-lives in vitro in synaptic
biochemical fractions76. ||Protein stability half-lives in vitro in total extracts84. ¶Protein stability half-lives in vivo (total mouse brain)83. #Gene name and relative
abundance within the neuropile transcriptome indicated by binned percentile ranks (for example, 0.8–0.9 indicates that a given mRNA is within the top 80–90% of
most abundant neuropile mRNAs (E.M.S., G. Tushev, I. Cajigas and T. Will, unpublished observations)).
after ER exit and whether local processing requires the presence of Golgi membranes45,50, these data strongly suggest that
neurons have evolved specialized means
to locally process membrane and secreted
proteins.
The lifetime of synaptic proteins is a critical determinant of how long, how far and at
what speed they can be stored and shipped.
Although long-lived proteins can potentially
be traded over long distances, high shipping
costs or slow transportation speed sometimes dictate local production. Analogously,
reliance on local production is likely for
proteins with a short lifetime and those that
cannot be rapidly recruited to a remote location. At present, however, the relative paucity
of data on the lifetime of synaptic receptors and other ion channels (TABLE 1) makes
it difficult to evaluate the contribution of
local membrane protein synthesis to basal
synaptic function and dendritic excitability.
However, it is clear that the levels of membrane proteins can be locally controlled in a
protein synthesis-dependent manner 47, and
evidence suggests that some forms of synaptic plasticity require a rapid, and probably
local, synthesis of specific receptors. More
specifically, two electrophysiological studies
have shown that acute (~25 minutes) and
specific blockade of AMPA receptor synthesis with small interfering RNA could impair
the expression of protein synthesis-dependent forms of synaptic plasticity 48,49. These
results thus indicate that new receptors can
be synthesized, assembled and trafficked to
synapses within 20–25 minutes in response
to exposure to a glutamate receptor agonist 48
or brain-derived neurotrophic factor 49. For
comparison, the complete progression of
nascent cargo through the secretory pathway
requires up to 120 minutes in specialized
secretory cells, such as pancreatic exocrine
cells46. Many secretory proteins are glycosylated and their sugars are modified as they
progress through the secretory pathway 51, a
feature that is commonly used to distinguish
mature and immature membrane proteins.
Indeed, studies in dissociated neurons suggest that nascent AMPA receptors dwell in
the secretory pathway for hours52, challenging the notion that new receptors can be
produced over tens of minutes. The reason
for this discrepancy is unknown. It is possible, however, that the glycosylation profiles
of mature proteins may differ for proteins
synthesized in the soma and those produced
locally in dendrites. Indeed, the overall low
abundance of bona fide Golgi membranes in
dendrites45, suggests that nascent dendritic
cargo may ‘bypass’ specific steps of secretory
processing 50 to allow for a more rapid delivery to the cell surface45. If this is the case,
then one expects that ion channels and other
membrane proteins will possess distinct
functional properties when synthesized in
the soma or in dendrites. This remains to be
determined.
NATURE REVIEWS | NEUROSCIENCE
Protein exchanges in the dendrite
It is now quite clear that, at least in dendritic spines, the local control of membrane
trafficking is a key mechanism to control
the level of postsynaptic neurotransmitter receptors53,54. However, we still do not
know much about the extent to which local
intracellular reservoirs exchange cargo with
other pools of receptors (for example, shaft
endosomes, the extrasynaptic membrane,
and so on) (FIG. 2c). In addition, whereas the
dynamics of receptors are now better understood at the scale of the individual synapse5,
little is known about receptor exchanges
over larger spatial scales. The rate of these
global exchanges must be different for proteins freely diffusing in the cytoplasm, those
in cellular membranes or those actively
transported in tubulovesicular structures
along microtubules. Owing to the saltatory nature of kinesin- or dynein-mediated
transport 55, it is easy to envision how microtubule-based transport could be controlled
to prevent (or enable) cargo exchange and
deposition over long distances. For example, phosphorylation of a kinesin family
member by calcium/calmodulin-dependent
protein kinase II (CaMKII) results in the
release of cargo56. As synaptic activity controls CaMKII activity 57, this could provide a
mechanism by which synaptic activity terminates dendritic tubulovesicular trafficking, resulting in the local release of cargo at
activated synapses. Importantly, long-range
VOLUME 14 | SEPTEMBER 2013 | 643
© 2013 Macmillan Publishers Limited. All rights reserved
PERSPECTIVES
Table 2 | High-throughput measurements of global protein stability
Model (tissue or cell
type)
Method
Protein turnover half-life (t1/2)
Refs
Median
Range
Mean ± SEM
n (proteins)
(0*) 7.6–217 h
32.3 ± 3.3 h
52
106
43 min
3,751
107
Min–months
8.5 h
576
108
45 min–22.5 h
9 h
100
109
in vitro
Yeast
Metabolic labelling and MS
34.25 h
Yeast‡
Tagged-protein library and
immunoblotting
43 min
Human lung cancer cells
Metabolic labelling and MS
20.4 h
Human lung cancer cells
Tagged-protein library and
photobleaching
Human cervical cancer
cells§
Metabolic labelling and MS
35.5 h
8.1 h–5 months
40.5 ± 1 h
4,106
110
Mouse myoblasts§
Metabolic labelling and MS
43.2 h
10.5 h–18 months
68.6 ± 1 h
3,528
110
Cultured neurons
(total proteins)
Metabolic labelling and MS
4.2 days
5.1 days
2,802
84
Cultured neurons
(synaptic proteins)
Metabolic labelling and MS
3.7 days
4.1 days
191
84
1,716
83
in vivo
Mouse (brain, liver and
blood)
Metabolic labelling and MS
0.17 min–11.5 months
Mouse (brain)
Metabolic labelling and MS
9 days
1,010
83
Mouse (liver)
Metabolic labelling and MS
3.5 days
1,122
83
Mouse (blood)
Metabolic labelling and MS
3 days
334
83
Mouse (liver)
Metabolic labelling and MS
4.2 days
14.4 h–9.8 days
4.1 ± 0.3 days
31
111
Mouse (kidney)
Metabolic labelling and MS
4.4 days
1.4 –12.4 days
4.9 ± 0.4 days
32
111
Mouse (cardiac muscle)
Metabolic labelling and MS
10.5 days
3.1 days–7.7 months
22.4 ± 7.6 days 30
111
Mouse (skeletal muscle)
Metabolic labelling and MS
28.3 days
3–53.3 days
27.3 ± 3.2 days 20
111
MS, mass spectrometry. *Zero decay after correction for protein dilution due to cell growth over a 51‑hour time course. ‡Protein synthesis was inhibited during the
experiment. §Non-dividing cells.
exchanges throughout dendrites also seem
to be limited for proteins diffusing in continuous structures such as the cytoplasm58,
the ER25 and the plasma membrane59. An
interesting notion is that dendritic spines
and PSDs act as diffusion sinks and traps
that — passively or by actual binding —
limit the exploration of large areas of dendritic terrain59 (FIG. 2c). Consistently, and in
contrast to local membrane trafficking, the
long-range diffusion of receptors at the cell
surface seems to make only a minor contribution to the fast turnover of relatively distal
synapses60 (but see REF. 61). Although the
entire extrasynaptic plasma membrane contains an enormous reserve pool of synaptic
receptors62–64, the relatively slow mobility of
membrane proteins5 and the overall complexity of the dendritic surface probably
limit the actual availability of nascent proteins at the dendritic surface. Consistently,
recent work suggests that anomalous diffusion within the ER restricts the dendritic
volume explored by nascent membrane
proteins before ER exit 25. Altogether, these
data indicate that there can be substantial
differences between local protein mobility
and the global exchange at the scale of the
entire neuron, and perhaps suggest that the
total levels of these proteins may not necessarily predict their local availability. In other
words, a local production of relatively longlived components in some instances may be
required to maintain a reliable supply.
Resource allocation in dendrites
The discovery of the rapid recruitment or
loss of proteins, such as AMPA receptors,
during long-term potentiation (LTP) and
long-term depression (LTD) of excitatory
synapses53,65–68 has proven to be a valuable
experimental paradigm for understanding
how basic cellular processes contribute to the
control of synaptic properties. The regulation
of protein exchanges between the postsynapse, the plasma membrane and intracellular
pools has emerged as a critical factor that
tunes synaptic strength and dendritic excitability 54,69. However, it is unclear what fraction of a specific protein is actually in use at
any given time and whether these fractions
are different for distinct types of proteins.
644 | SEPTEMBER 2013 | VOLUME 14
Despite various shapes and functional
properties, synapses display invariant features, resulting, for example, in a strong
correlation between the number of available synaptic vesicles and the number of
postsynaptic receptors or the PSD area70,71.
Exploiting such correlations to study ultrastructural changes occurring during synaptic
potentiation, recent experiments showed
that despite striking variations of synapse
size and density, total excitatory and inhibitory PSD areas per unit of dendritic length
are unchanged in activated dendrites72. This
strongly indicates that plasticity induction may require redistribution of limited
resources that are ‘shared’ between synapses
(FIG. 2c). Consistent with this idea, imaging
of PSD95 (also known as DLG4) dynamics
in vivo showed that neighbouring synapses
locally compete for available molecules
that are estimated to constitute ~1% of
the total protein content at the scale of the
entire cell73. As suggested by the observation that overexpression of PSD95 can lead
to increased synapse strength and size74, the
basal levels of the endogenous protein seem
www.nature.com/reviews/neuro
© 2013 Macmillan Publishers Limited. All rights reserved
PERSPECTIVES
to be limiting. It is thus tempting to speculate that the wealth of activity-dependent
mechanisms that facilitate PSD95 recruitment at synapses75 may lead to an abrupt
and local depletion of available resources.
The high abundance of Psd95 mRNAs
among dendritic RNAs suggests an important local synthesis of the protein4 (TABLE 1),
which would thus be crucial for maintaining the local availability of this otherwise
relatively stable76,77 and abundant protein.
Global versus local protein lifetimes
Measurements of apparent protein stability
in PSD fractions revealed half-lives, ranging
from minutes up to tens of hours76 (TABLE 1).
If it is pretty clear why and how cytoplasmic
proteins with short lifetimes might be locally
synthesized at synapses78 (FIG. 2c), the case of
surface ion channels is more problematic. As
receptor dwell-times at synapses and internalization rates are on the order of seconds
and minutes, respectively 5,79,80, receptors
undergo many rounds of internalization and
recycling before being degraded, a process
that is likely to be all the more efficient in
dendritic spines54. This relatively fast recycling process may allow a local concentration
of receptors taken from a diffuse pool, perhaps obviating the need for local synthesis.
Conversely, the activity-dependent targeting of internalized receptors for lysosomal
degradation79 and the activity-dependent
recruitment of the proteasome in dendritic
spines81 suggest that synaptic protein lifetime is subject to potent and compartmentspecific regulations76,82. Indeed, compared
with average protein half-lives measured in
bacteria, fibroblasts or total cultured neuron
extracts83,84 (TABLES 1,2), values obtained
from isolated synaptic fractions suggest
that synaptic proteins have a fast turnover
(mean half-life ± SEM = 3.3 ± 0.5 hours for
global radioactivity loss after metabolic
labelling, n = 6 experiments, and mean halflife ± SEM = 9 ± 3 hours for the 10 proteins
for which it has specifically been measured;
see REF. 76 and TABLE 1). More specifically,
measurements of the half-life of PSD95 in
total neuronal extracts (~36–87 hours)77,84
and synaptic fractions (~7 hours)76 revealed
sensible differences between the global and
local stability of this synaptic protein.
Interestingly, data suggest that the
polyubiquitylation and degradation of a few
‘master’ scaffolding proteins such as SAP90/
PSD95-associated protein 1 (SAPAP1; also
known as GKAP and DLGAP1) and SH3
and multiple ankyrin repeat domains protein
(SHANK) by the proteasome leads to the
co‑depletion of other synaptic molecules (for
example, GluN2B) that are not thought to
be targeted by the proteasome76. Although
it is not yet clear whether these proteins are
locally degraded or simply relocalized to
extrasynaptic pools, these results suggest that
synaptic proteins live ‘in the fast lane’ and
may have a shortened lifetime. Consistent
with this idea, the maintenance of dendritic
structure seems to require ongoing secretory
trafficking 26, indicating that key membrane
or secreted proteins need to be replaced
within hours. Combined with the overall low
mobility of integral membrane proteins at the
scale of an entire dendrite, variations of their
stability in space and time may be such that
an acute and local synthesis of ion channels
may be necessary. Similarly, and as discussed
above for PSD95, the efficient sequestration
of available cytoplasmic proteins, as well as
the rapid and input-specific variations of
available binding slots and the local competition between synapses, may result in the local
depletion of available molecules and may
thus require local synthesis.
PrP products and synaptic tagging
It is commonly believed that although
changes in the phosphorylation status of
receptors and associated molecules and their
local trafficking can drive plasticity of postsynaptic responses over the course of minutes53,54, longer lasting changes in synaptic
strength requires the activation of transcriptional programmes and/or de novo protein
synthesis4. A common view is that the resulting ‘plasticity-related protein’ (PrP) products
are released in a shared pool but only incorporated in those synapses that have been
tagged by (during) their prior potentiation85,
explaining how resources produced remotely
can be locally incorporated in an inputspecific manner. The actual availability of
this common pool, the identity and distinctness of PrP products remain unknown. A
recent paper described elegant approaches
to tackle these questions. By combining focal
glutamate uncaging and pharmacological
activation of protein kinase A, Govindarajan
et al.7 controlled the induction of ‘early’
or ‘late’ forms of LTP (E-LTP and L‑LTP,
respectively) at individual synapses (FIG. 4a).
Consistent with the original formulation of
the synaptic tagging and capture model85,
the authors showed that PrP products
produced in response to L‑LTP induction
and recruited at activated synapses can
also be recruited by nearby synapses previously or subsequently tagged by E‑LTP
induction, converting E‑LTP to L‑LTP at
tagged synapses7 (FIG. 4b). Taking advantage
of this temporal bidirectionality and the
NATURE REVIEWS | NEUROSCIENCE
synthesis-dependent or synthesis-independent nature of PrP products and the synaptic
tag, the authors selectively measured the
half-lives of these components (~90 minutes
for PrP products and ~120 minutes for the
tag). By monitoring L‑LTP and E-LTP-toL‑LTP conversion in synapse pairs located
on the same or distinct dendritic branches,
they found that the propagation of PrP
products was limited by branch points
(FIG. 4c), providing a compelling illustration
of the influence of cellular morphology on
resource sharing in elaborate dendrites.
By focusing on specific forms of synaptic
plasticity that rely on de novo synthesis of
cytoplasmic78 versus membrane proteins48,49,
it should be now possible to provide a more
complete picture of how these temporal and
morphological constraints shape resource
allocation throughout dendrites. In the long
run, it will be interesting to determine how
differences in protein mobility and stability,
specific dendritic morphologies and the partition between somatic and dendritic protein
synthesis determine how fast, for how long
and to what extent synaptic properties can
be adjusted in an input-specific manner.
Concluding remarks
Although it is now well established that
various forms of synaptic plasticity require
de novo protein synthesis4, the identity of
the proteins that are being synthesized still
has to be determined. Conversely, evidence
suggests that activity-dependent protein
degradation is also tightly controlled during
synaptic plasticity 76,81. However, plasticityrelated ‘degromes’ have not been characterized. Future characterization of both the
plasticity-related proteome and degrome
must be quantitative, as it is likely that the
major plasticity-related changes will be
in the relative stoichiometries rather than
identities of synaptic proteins. Although the
standard experimental method to establish
the protein synthesis- and degradationdependence of a given biological response
is still through the use of generic inhibitors
of the translation and degradation machineries, the development and availability of
experimental means to selectively inhibit the
synthesis of specific proteins and trigger or
block their degradation is important.
As illustrated by TABLE 1, there is growing information but limited consensus on
the global turnover of synaptic proteins and
dendritic ion channels (to say nothing about
mRNA processing and translation) to start
mining for probable links between these
parameters. Although available data suggest
important discrepancies between protein
VOLUME 14 | SEPTEMBER 2013 | 645
© 2013 Macmillan Publishers Limited. All rights reserved
PERSPECTIVES
a
L-LTP
or
E-LTP
E-LTP
1
0
Normal spine
volume
Normal spine
volume
L-LTP
2
+ANI
–25 25 75 125 175 225
2
1
+ANI
0
–25 25 75 125 175 225
Time (min)
L-LTP
2
L-LTP
E-LTP
1
0
+ANI
E-LTP
Normal spine
volume
b
Normal spine
volume
Time (min)
–25 25 75 125 175 225
2
1
0
L-LTP
2
1
0
–25 25 75 125 175 225
Time (min)
+ANI
+ANI
Normal spine
volume
E-LTP
Normal spine
volume
Time (min)
–25 25 75 125 175 225
2
1
0
Time (min)
S2'
S1 L-LTP
S2 E-LTP
–25 25 75 125 175 225
Time (min)
Normalized S2 or S2' volume at
100 min after L-LTP induction
c
+ANI
+ANI
2
stability measurements taken from cultured
cells and in vivo by mass spectrometry after
metabolic labelling (TABLE 2), it is not yet clear
to what extent these differences are real or
owing to differences in technique. Indeed,
important differences may exist in the overall
accessibility and recycling of labelled precursors in distinct organs and cell populations
in vivo, which complicates the interpretations
of available results. Importantly, the contribution of local ongoing protein synthesis to dendritic protein homeostasis and hence synaptic
properties and dendritic excitability and its
deregulation in pathological contexts is most
likely to be different in distinct neuronal types
and specific compartments (for example, dendritic spines). It is thus now critical to develop
appropriate tools to selectively label, detect
and purify mRNA and proteins in targeted
cell populations and cell compartments in
normal and pathological brains. Although
experimental strategies to label RNAs86 and
proteins87 in a genetically restricted manner
exist and have been characterized in cultured
cells, efforts should now be made to generate
and characterize genetically modified models
to use these molecular tools in vivo.
Cyril Hanus and Erin M. Schuman are at The Max
Planck Institute for Brain Research, Max von Laue
Strasse 4, 60438 Frankfurt, Germany.
1.5
1
r = 0.4
0.5
0
r = 0.96
Correspondence to E.M.S. e-mail: [email protected]
Same branch
Sister branch
0
20
40
60
80
Distance between S1 and S2 or S2' (µm)
Figure 4 | Synaptic competition and resource allocation in dendrites. a | Early long-term
potentiation (E-LTP; red circle) and late LTP (L-LTP; blue circle) and the
resulting
spine
enlargement
Nature
Reviews
| Neuroscience
(red shading around spine; left panel) can be induced by focal glutamate uncaging at individual
synapses (indicated by arrows in the graphs), and their contrasting reliance on protein synthesis is
shown by exposure to the translation inhibitor anisomycin (ANI) (middle and right panels). Whereas
L‑LTP induction results in a long-lasting increase in spine volume (middle panel, dark blue trace) that
is completely abolished by protein synthesis inhibitors (light blue trace), E‑LTP induction results in
a more transient spine enlargement (right panel, dark red trace) that is not affected by ANI (light red
trace). b | Synapse tagging and capture in pairs of spines activated by L‑LTP (‘donor’ synapse) or
E‑LTP (‘acceptor’ synapse) is demonstrated by the conversion of E‑LTP to L-LTP at ‘tagged’ synapses
stimulated after L‑LTP induction (left scheme) or before L‑LTP induction (right scheme). In the left
scheme, top graph, the conversion of E-LTP to L‑LTP (shown by the lack of decay in the E-LTP trace)
is not blocked by ANI and thus relies on the recruitment of proteins (synaptic tags) induced after
prior L‑LTP induction. By contrast, ANI added concurrently with the L‑LTP induction protocol prevents L‑LTP expression at both synapses, but E-LTP induction is unaffected (bottom graph). In the
right scheme, induction of E-LTP followed by L‑LTP induction at a neighbouring synapse allows the
conversion of E‑LTP to L‑LTP (indicated by lack of decay of the red trace in the top graph).
Application of ANI concurrently with E-LTP induction results in the usual E-LTP formation and decay
profile (bottom graph, red trace), indicating that it has not been converted to L-LTP. ANI applied
concurrently with the L-LTP induction protocol results in a complete block of spine volume changes
(bottom graph) . c | The experiment described in part b, in which prior L-LTP induction allows E-LTPto-L-LTP conversion in tagged synapses, is used here to test the distance of propagation of the
plasticity-related protein products (synaptic tags) in synapse pairs at varying distances in the same
(S2) or sister (S2’) dendritic branches. The graph on the right shows that the propagation of synaptic
tag proteins is reduced in a sister branch compared with the same branch, and illustrates how dendritic branch points restrict the distribution of these proteins. Figure is adapted, with permission,
from REF. 7 © (2011) Elsevier.
646 | SEPTEMBER 2013 | VOLUME 14
doi:10.1038/nrn3546
Published online 31 July 2013
Ishizuka, N., Cowan, W. M. & Amaral, D. G.
A quantitative analysis of the dendritic organization of
pyramidal cells in the rat hippocampus. J. Comp.
Neurol. 362, 17–45 (1995).
2. Ascoli, G. A., Donohue, D. E. & Halavi, M.
NeuroMorpho.Org: a central resource for neuronal
morphologies. J. Neurosci. 27, 9247–9251 (2007).
3. Moran, U., Phillips, R. & Milo, R. SnapShot: key
numbers in biology. Cell 141, 1262–1262.e1
(2010).
4. Cajigas, I. J., Will, T. & Schuman, E. M. Protein
homeostasis and synaptic plasticity. EMBO J. 29,
2746–2752 (2010).
5. Triller, A. & Choquet, D. Surface trafficking of receptors
between synaptic and extrasynaptic membranes: and
yet they do move! Trends Neurosci. 28, 133–139
(2005).
6. Nusser, Z. Differential subcellular distribution of ion
channels and the diversity of neuronal function. Curr.
Opin. Neurobiol. 22, 366–371 (2012).
7. Govindarajan, A., Israely, I., Huang, S. Y. &
Tonegawa, S. The dendritic branch is the preferred
integrative unit for protein synthesis-dependent LTP.
Neuron 69, 132–146 (2011).
8. Bassell, G. J. & Warren, S. T. Fragile X syndrome: loss
of local mRNA regulation alters synaptic development
and function. Neuron 60, 201–214 (2008).
9. Vo, N. K., Cambronne, X. A. & Goodman, R. H.
MicroRNA pathways in neural development and
plasticity. Curr. Opin. Neurobiol. 20, 457–465 (2010).
10. Lin, A. C. & Holt, C. E. Local translation and directional
steering in axons. EMBO J. 26, 3729–3736 (2007).
11. Mabb, A. M. & Ehlers, M. D. Ubiquitination in
postsynaptic function and plasticity. Annu. Rev. Cell
Dev. Biol. 26, 179–210 (2010).
12.Golgi, C. Sulla Struttura Delle Fibre Nervose Midollate
Periferiche e Centrali, Ricerche del C. Golgi, Lette al
R. Istituto Lombardo Nell’ Adunanza del Dicembre 18
Dicembre 1879 (Milano, 1880) (in Italian).
1.
www.nature.com/reviews/neuro
© 2013 Macmillan Publishers Limited. All rights reserved
PERSPECTIVES
13. Ramón y Cajal, S. Histologie du Système Nerveux de
l’Homme & des Vertébrés (A. Maloine, 1909) (in
French).
14. Kuo, C. T., Zhu, S., Younger, S., Jan, L. Y. & Jan, Y. N.
Identification of E2/E3 ubiquitinating enzymes and
caspase activity regulating Drosophila sensory neuron
dendrite pruning. Neuron 51, 283–290 (2006).
15.Kim, A. H. et al. A centrosomal Cdc20–APC pathway
controls dendrite morphogenesis in postmitotic
neurons. Cell 136, 322–336 (2009).
16. Huang, Z., Zang, K. & Reichardt, L. F. The origin
recognition core complex regulates dendrite and spine
development in postmitotic neurons. J. Cell Biol. 170,
527–535 (2005).
17. Baas, P. W., Deitch, J. S., Black, M. M. & Banker, G. A.
Polarity orientation of microtubules in hippocampal
neurons: uniformity in the axon and nonuniformity in
the dendrite. Proc. Natl Acad. Sci. USA 85,
8335–8339 (1988).
18.Stiess, M. et al. Axon extension occurs independently
of centrosomal microtubule nucleation. Science 327,
704–707 (2010).
19. Lin, S., Liu, M., Mozgova, O. I., Yu, W. & Baas, P. W.
Mitotic motors coregulate microtubule patterns in
axons and dendrites. J. Neurosci. 32, 14033–14049
(2012).
20. Herculano-Houzel, S. Scaling of brain metabolism with
a fixed energy budget per neuron: implications for
neuronal activity, plasticity and evolution. PLoS ONE
6, e17514 (2011).
21. Ostroff, L. E., Fiala, J. C., Allwardt, B. & Harris, K. M.
Polyribosomes redistribute from dendritic shafts into
spines with enlarged synapses during LTP in
developing rat hippocampal slices. Neuron 35,
535–545 (2002).
22. Bourne, J. N., Sorra, K. E., Hurlburt, J. & Harris, K. M.
Polyribosomes are increased in spines of CA1
dendrites 2 h after the induction of LTP in mature rat
hippocampal slices. Hippocampus 17, 1–4 (2007).
23.Griffiths, G. et al. The dynamic nature of the Golgi
complex. J. Cell Biol. 108, 277–297 (1989).
24. Horton, A. C. & Ehlers, M. D. Dual modes of
endoplasmic reticulum-to‑Golgi transport in dendrites
revealed by live-cell imaging. J. Neurosci. 23,
6188–6199 (2003).
25.Cui-Wang, T. et al. Local zones of endoplasmic
reticulum complexity confine cargo in neuronal
dendrites. Cell 148, 309–321 (2012).
26.Horton, A. C. et al. Polarized secretory trafficking
directs cargo for asymmetric dendrite growth and
morphogenesis. Neuron 48, 757–771 (2005).
27.Ye, B. et al. Growing dendrites and axons differ in their
reliance on the secretory pathway. Cell 130, 717–729
(2007).
28. Sheng, M. & Hoogenraad, C. C. The postsynaptic
architecture of excitatory synapses: a more
quantitative view. Annu. Rev. Biochem. 76, 823–847
(2007).
29.Pielot, R. et al. SynProt: a database for proteins of
detergent-resistant synaptic protein preparations.
Front. Synaptic Neurosci. 4, 1 (2012).
30.Cheng, D. et al. Relative and absolute quantification of
postsynaptic density proteome isolated from rat
forebrain and cerebellum. Mol. Cell. Proteomics 5,
1158–1170 (2006).
31.Chen, X. et al. Mass of the postsynaptic density and
enumeration of three key molecules. Proc. Natl Acad.
Sci. USA 102, 11551–11556 (2005).
32. Grutzendler, J., Kasthuri, N. & Gan, W. B. Long-term
dendritic spine stability in the adult cortex. Nature
420, 812–816 (2002).
33.Trachtenberg, J. T. et al. Long-term in vivo imaging of
experience-dependent synaptic plasticity in adult
cortex. Nature 420, 788–794 (2002).
34. Holcman, D. & Triller, A. Modeling synaptic dynamics
driven by receptor lateral diffusion. Biophys. J. 91,
2405–2415 (2006).
35.Czondor, K. et al. Unified quantitative model of AMPA
receptor trafficking at synapses. Proc. Natl Acad. Sci.
USA 109, 3522–3527 (2012).
36. Poon, M. M., Choi, S. H., Jamieson, C. A.,
Geschwind, D. H. & Martin, K. C. Identification of
process-localized mRNAs from cultured rodent
hippocampal neurons. J. Neurosci. 26,
13390–13399 (2006).
37. Zhong, J., Zhang, T. & Bloch, L. M. Dendritic mRNAs
encode diversified functionalities in hippocampal
pyramidal neurons. BMC Neurosci. 7, 17 (2006).
38.Lein, E. S. et al. Genome-wide atlas of gene expression
in the adult mouse brain. Nature 445, 168–176
(2007).
39.Cajigas, I. J. et al. The local transcriptome in the
synaptic neuropil revealed by deep sequencing and
high-resolution imaging. Neuron 74, 453–466 (2012).
40.Kamme, F. et al. Single-cell microarray analysis in
hippocampus CA1: demonstration and validation of
cellular heterogeneity. J. Neurosci. 23, 3607–3615
(2003).
41. Fiala, J. C. & Harris, K. M. Dendrite Structure.
Dendrites (Oxford Univ. Press, 1999).
42.Shepard, P. J. et al. Complex and dynamic landscape
of RNA polyadenylation revealed by PAS-Seq. RNA
17, 761–772 (2011).
43.Derti, A. et al. A quantitative atlas of polyadenylation in
five mammals. Genome Res. 22, 1173–1183 (2012).
44. Kanehisa, M., Goto, S., Sato, Y., Furumichi, M. &
Tanabe, M. KEGG for integration and interpretation of
large-scale molecular data sets. Nucleic Acids Res. 40,
D109–D114 (2012).
45. Hanus, C. & Ehlers, M. D. Secretory outposts for the
local processing of membrane cargo in neuronal
dendrites. Traffic 9, 1437–1445 (2008).
46. Palade, G. Intracellular aspects of the process of
protein synthesis. Science 189, 867 (1975).
47.Sutton, M. A. et al. Miniature neurotransmission
stabilizes synaptic function via tonic suppression of
local dendritic protein synthesis. Cell 125, 785–799
(2006).
48. Mameli, M., Balland, B., Luján, R. & Lüscher, C. Rapid
synthesis and synaptic insertion of GluR2 for mGluRLTD in the ventral tegmental area. Science 317,
530–533 (2007).
49.Fortin, D. A. et al. Brain-derived neurotrophic factor
activation of CaM-kinase kinase via transient receptor
potential canonical channels induces the translation
and synaptic incorporation of GluA1‑containing
calcium-permeable AMPA receptors. J. Neurosci. 32,
8127–8137 (2012).
50.Jeyifous, O. et al. SAP97 and CASK mediate sorting of
NMDA receptors through a previously unknown
secretory pathway. Nature Neurosci. 12, 1011–1019
(2009).
51. Moremen, K. W., Tiemeyer, M. & Nairn, A. V.
Vertebrate protein glycosylation: diversity, synthesis
and function. Nature Rev. Mol. Cell Biol. 13, 448–462
(2012).
52. Greger, I. H., Khatri, L. & Ziff, E. B. RNA editing at
Arg607 controls AMPA receptor exit from the
endoplasmic reticulum. Neuron 34, 759–772 (2002).
53. Collingridge, G. L., Isaac, J. T. & Wang, Y. T. Receptor
trafficking and synaptic plasticity. Nature Rev.
Neurosci. 5, 952–962 (2004).
54. Newpher, T. M. & Ehlers, M. D. Glutamate receptor
dynamics in dendritic microdomains. Neuron 58,
472–497 (2008).
55.Derr, N. D. et al. Tug‑of‑war in motor protein
ensembles revealed with a programmable DNA
origami scaffold. Science 338, 662–665 (2012).
56. Guillaud, L., Wong, R. & Hirokawa, N. Disruption of
KIF17–Mint1 interaction by CaMKII-dependent
phosphorylation: a molecular model of kinesin-cargo
release. Nature Cell Biol. 10, 19–29 (2008).
57. Lisman, J., Yasuda, R. & Raghavachari, S. Mechanisms
of CaMKII action in long-term potentiation. Nature
Rev. Neurosci. 13, 169–182 (2012).
58. Santamaria, F., Wils, S., De Schutter, E. &
Augustine, G. J. Anomalous diffusion in Purkinje cell
dendrites caused by spines. Neuron 52, 635–648
(2006).
59. Bressloff, P. C. & Earnshaw, B. A. Diffusion-trapping
model of receptor trafficking in dendrites. Phys. Rev. E
Stat. Nonlin. Soft Matter Phys. 75, 041915 (2007).
60.Petrini, E. M. et al. Endocytic trafficking and recycling
maintain a pool of mobile surface AMPA receptors
required for synaptic potentiation. Neuron 63,
92–105 (2009).
61. Adesnik, H., Nicoll, R. A. & England, P. M.
Photoinactivation of native AMPA receptors reveals
their real-time trafficking. Neuron 48, 977–985
(2005).
62. Tovar, K. R. & Westbrook, G. L. Mobile NMDA
receptors at hippocampal synapses. Neuron 34,
255–264 (2002).
63. Thomas, P., Mortensen, M., Hosie, A. M. &
Smart, T. G. Dynamic mobility of functional GABAA
receptors at inhibitory synapses. Nature Neurosci. 8,
889–897 (2005).
64.Kasugai, Y. et al. Quantitative localisation of synaptic
and extrasynaptic GABAA receptor subunits on
hippocampal pyramidal cells by freeze-fracture replica
immunolabelling. Eur. J. Neurosci. 32, 1868–1888
(2010).
NATURE REVIEWS | NEUROSCIENCE
65.Luthi, A. et al. Hippocampal LTD expression involves a
pool of AMPARs regulated by the NSF–GluR2
interaction. Neuron 24, 389–399 (1999).
66.Hayashi, Y. et al. Driving AMPA receptors into
synapses by LTP and CaMKII: requirement for GluR1
and PDZ domain interaction. Science 287,
2262–2267 (2000).
67.Lu, W. et al. Activation of synaptic NMDA receptors
induces membrane insertion of new AMPA receptors
and LTP in cultured hippocampal neurons. Neuron 29,
243–254 (2001).
68. Malinow, R. & Malenka, R. C. AMPA receptor
trafficking and synaptic plasticity. Annu. Rev. Neurosci.
25, 103–126 (2002).
69. Lai, H. C. & Jan, L. Y. The distribution and targeting of
neuronal voltage-gated ion channels. Nature Rev.
Neurosci. 7, 548–562 (2006).
70. Harris, K. M. & Stevens, J. K. Dendritic spines of CA 1
pyramidal cells in the rat hippocampus: serial electron
microscopy with reference to their biophysical
characteristics. J. Neurosci. 9, 2982–2997 (1989).
71. Masugi-Tokita, M. & Shigemoto, R. High-resolution
quantitative visualization of glutamate and GABA
receptors at central synapses. Curr. Opin. Neurobiol.
17, 387–393 (2007).
72. Bourne, J. N. & Harris, K. M. Coordination of size and
number of excitatory and inhibitory synapses results
in a balanced structural plasticity along mature
hippocampal CA1 dendrites during LTP. Hippocampus
21, 354–373 (2011).
73. Gray, N. W., Weimer, R. M., Bureau, I. & Svoboda, K.
Rapid redistribution of synaptic PSD‑95 in the
neocortex in vivo. PLoS Biol. 4, e370 (2006).
74. El‑Husseini, A. E., Schnell, E., Chetkovich, D. M.,
Nicoll, R. A. & Bredt, D. S. PSD‑95 involvement in
maturation of excitatory synapses. Science 290,
1364–1368 (2000).
75. Keith, D. & El‑Husseini, A. Excitation control:
balancing PSD‑95 function at the synapse. Front. Mol.
Neurosci. 1, 4 (2008).
76. Ehlers, M. D. Activity level controls postsynaptic
composition and signaling via the ubiquitin-proteasome
system. Nature Neurosci. 6, 231–242 (2003).
77. El‑Husseini Ael, D. et al. Synaptic strength regulated
by palmitate cycling on PSD‑95. Cell 108, 849–863
(2002).
78. Waung, M. W., Pfeiffer, B. E., Nosyreva, E. D.,
Ronesi, J. A. & Huber, K. M. Rapid translation of Arc/
Arg3.1 selectively mediates mGluR-dependent LTD
through persistent increases in AMPAR endocytosis
rate. Neuron 59, 84–97 (2008).
79. Ehlers, M. D. Reinsertion or degradation of AMPA
receptors determined by activity-dependent endocytic
sorting. Neuron 28, 511–525 (2000).
80.Lin, J. W. et al. Distinct molecular mechanisms and
divergent endocytotic pathways of AMPA receptor
internalization. Nature Neurosci. 3, 1282–1290
(2000).
81. Bingol, B. & Schuman, E. M. Activity-dependent
dynamics and sequestration of proteasomes in
dendritic spines. Nature 441, 1144–1148 (2006).
82. Patrick, G. N., Bingol, B., Weld, H. A. &
Schuman, E. M. Ubiquitin-mediated proteasome
activity is required for agonist-induced endocytosis of
GluRs. Curr. Biol. 13, 2073–2081 (2003).
83. Price, J. C., Guan, S., Burlingame, A., Prusiner, S. B. &
Ghaemmaghami, S. Analysis of proteome dynamics in
the mouse brain. Proc. Natl Acad. Sci. USA 107,
14508–14513 (2010).
84.Cohen, L. D. et al. Metabolic turnover of synaptic
proteins: kinetics, interdependencies and implications
for synaptic maintenance. PLoS ONE 8, e63191
(2013).
85. Frey, U. & Morris, R. G. Synaptic tagging and longterm potentiation. Nature 385, 533–536 (1997).
86. Cleary, M. D., Meiering, C. D., Jan, E., Guymon, R. &
Boothroyd, J. C. Biosynthetic labeling of RNA with
uracil phosphoribosyltransferase allows cell-specific
microarray analysis of mRNA synthesis and decay.
Nature Biotech. 23, 232–237 (2005).
87.Ngo, J. T. et al. Cell-selective metabolic labeling of
proteins. Nature Chem. Biol. 5, 715–717 (2009).
88. Ramón y Cajal, S. Les Nouvelles Idées sur la
Structure du Système Nerveux Chez l’Homme et
Chez les Vertébrés (C. Reinwald, 1894) (in French).
89. Claiborne, B. J., Amaral, D. G. & Cowan, W. M.
Quantitative, three-dimensional analysis of granule
cell dendrites in the rat dentate gyrus. J. Comp.
Neurol. 302, 206–219 (1990).
90.Andersen, P. The Hippocampus Book (Oxford Univ.
Press, 2007).
VOLUME 14 | SEPTEMBER 2013 | 647
© 2013 Macmillan Publishers Limited. All rights reserved
PERSPECTIVES
91. Sharma, K., Fong, D. K. & Craig, A. M. Postsynaptic
protein mobility in dendritic spines: long-term
regulation by synaptic NMDA receptor activation.
Mol. Cell. Neurosci. 31, 702–712 (2006).
92. Kerr, J. M. & Blanpied, T. A. Subsynaptic AMPA
receptor distribution is acutely regulated by actindriven reorganization of the postsynaptic density.
J. Neurosci. 32, 658–673 (2012).
93. Ashby, M. C., Maier, S. R., Nishimune, A. &
Henley, J. M. Lateral diffusion drives constitutive
exchange of AMPA receptors at dendritic spines and is
regulated by spine morphology. J. Neurosci. 26,
7046–7055 (2006).
94. Bats, C., Groc, L. & Choquet, D. The interaction
between Stargazin and PSD‑95 regulates AMPA
receptor surface trafficking. Neuron 53, 719–734
(2007).
95. Serge, A., Fourgeaud, L., Hemar, A. & Choquet, D.
Receptor activation and homer differentially control
the lateral mobility of metabotropic glutamate
receptor 5 in the neuronal membrane. J. Neurosci.
22, 3910–3920 (2002).
96.Jacob, T. C. et al. Gephyrin regulates the cell surface
dynamics of synaptic GABAA receptors. J. Neurosci.
25, 10469–10478 (2005).
97. Renner, M., Schweizer, C., Bannai, H., Triller, A. &
Levi, S. Diffusion barriers constrain receptors at
synapses. PLoS ONE 7, e43032 (2012).
98. Okabe, S., Urushido, T., Konno, D., Okado, H. &
Sobue, K. Rapid redistribution of the postsynaptic
density protein PSD‑Zip45 (Homer 1c) and its
differential regulation by NMDA receptors and calcium
channels. J. Neurosci. 21, 9561–9571 (2001).
99. Nakagawa, T., Engler, J. A. & Sheng, M. The dynamic
turnover and functional roles of α-actinin in dendritic
spines. Neuropharmacology 47, 734–745 (2004).
100.Kuriu, T., Inoue, A., Bito, H., Sobue, K. & Okabe, S.
Differential control of postsynaptic density scaffolds
via actin-dependent and -independent mechanisms.
J. Neurosci. 26, 7693–7706 (2006).
101. Zheng, C. Y., Petralia, R. S., Wang, Y. X., Kachar, B. &
Wenthold, R. J. SAP102 is a highly mobile MAGUK in
spines. J. Neurosci. 30, 4757–4766 (2010).
102.Nakagawa, T. et al. Quaternary structure, protein
dynamics, and synaptic function of SAP97 controlled by
L27 domain interactions. Neuron 44, 453–467 (2004).
103.Tai, C. Y., Mysore, S. P., Chiu, C. & Schuman, E. M.
Activity-regulated N‑cadherin endocytosis. Neuron 54,
771–785 (2007).
104.Star, E. N., Kwiatkowski, D. J. & Murthy, V. N. Rapid
turnover of actin in dendritic spines and its regulation
by activity. Nature Neurosci. 5, 239–246 (2002).
105.Charrier, C. et al. A crosstalk between β1 and β3
integrins controls glycine receptor and gephyrin
trafficking at synapses. Nature Neurosci. 13,
1388–1395 (2010).
106.Pratt, J. M. et al. Dynamics of protein turnover, a
missing dimension in proteomics. Mol. Cell Proteom.
1, 579–591 (2002).
107.Ghaemmaghami, S. et al. Global analysis of protein
expression in yeast. Nature 425, 737–741 (2003).
108.Doherty, M. K., Hammond, D. E., Clague, M. J.,
Gaskell, S. J. & Beynon, R. J. Turnover of the human
proteome: determination of protein intracellular
stability by dynamic SILAC. J. Proteome Res. 8,
104–112 (2009).
648 | SEPTEMBER 2013 | VOLUME 14
109.Eden, E. et al. Proteome half-life dynamics in living
human cells. Science 331, 764–768 (2011).
110.Cambridge, S. B. et al. Systems-wide proteomic
analysis in mammalian cells reveals conserved,
functional protein turnover. J. Proteome Res. 10,
5275–5284 (2011).
111. Claydon, A. J., Thom, M. D., Hurst, J. L. &
Beynon, R. J. Protein turnover: measurement of
proteome dynamics by whole animal metabolic
labelling with stable isotope labelled amino acids.
Proteomics 12, 1194–1206 (2012).
Acknowledgments
We are particularly grateful to L. Kochen for the data shown
in figure 2 and G. Tushev for his help with data analysis. We
thank I. Cajigas, S. tom Dieck and G. Tushev for their critical
reading of the manuscript. We apologize to those whose
work we could not cite because of space limitations. Work
in the laboratory of E.M.S. is supported by the Max Planck
Society, the European Research Council, DFG CRC 902 and
10 8 0 , a n d t h e D F G C l u s t e r o f E x c e l l e n c e f o r
Macromolecular Complexes. C.H. is supported by a Marie
Curie career integration grant.
Competing interests statement
The authors declare no competing financial interests.
FURTHER INFORMATION
KEGG: http://www.genome.jp/kegg/
ALL LINKS ARE ACTIVE IN THE ONLINE PDF
www.nature.com/reviews/neuro
© 2013 Macmillan Publishers Limited. All rights reserved